ClawHub

Eureka Feedback

Security checks across malware telemetry and agentic risk

Overview

This skill is transparent about contacting Eureka, but it can send user or task details to a Telegram-linked agent and a fixed reply recipient.

Install only if you intend to use this specific Eureka and Telegram workflow. Before use, verify that `clawdbot`, `@Eureka_Agent_Bot`, the `mason` reply account, and the numeric recipient ID belong to you or a trusted operator. Do not send secrets, credentials, private project data, or personal information unless you have confirmed where the messages and replies will go.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger phrase "ask eureka" is broad, natural-language-like, and likely to collide with ordinary conversation. In a skill that forwards messages to another agent, accidental activation can cause unintended delegation or disclosure of user content to an external channel.

Vague Triggers

Medium
Confidence
86% confidence
Finding
The trigger "check with eureka" is ambiguous and resembles common conversational phrasing, making unintentional invocation plausible. Because this skill is designed to contact a separate agent over Telegram tooling, mistaken activation can leak task details or user data outside the current agent boundary.

Shadow Command Trigger

Medium
Category
Trigger Abuse
Confidence
78% confidence
Finding
The phrase "ask eureka" overlaps with the built-in command namespace around "ask," creating a shadowing/confusion risk. This can cause the wrong handler to run, potentially redirecting user intent into a skill that sends content to another agent, increasing the chance of unintended data transfer or command execution flow changes.

VirusTotal

58/58 vendors flagged this skill as clean.

View on VirusTotal